Undertreatment of pain is common even when caused by serious illness. We examined whether physician–patient communication (particularly language indicating physician certainty) was associated with incomplete (i.e., premature closure) of pain assessment among patients with serious illness. Standardized patients (SPs) trained to portray patients with serious illness conducted unannounced, covertly audio-recorded visits to 20 consenting family physicians and 20 medical specialists. We coded extent of pain assessment, physician voice tone, and a measure of the degree to which physicians explored and validated patient concerns. To assess physician certainty, we searched transcripts for use of words that conveyed certainty using the Linguistic Inquiry and Word Count program. SP role fidelity was 94%, and few physicians were suspicious that they had seen an SP (14% of visits). Regression analyses showed that physicians who used more certainty language engaged in less thorough assessment of pain (β = –0.48, p < .05). Conversely, physicians who engaged in more exploring and validating of patient concerns (β = 0.27, p < .05) had higher ratings on anxiety/concerned voice tone (β = 0.25, p <.01) and engaged in more thorough assessment of pain. Together, these three factors accounted for 38% of the variance in pain assessment. Physicians who convey certainty in discussions with patients suffering from pain may be more likely to close prematurely their assessment of pain. We found that expressions of physician concern and responsiveness (curiosity) were associated with superior pain assessment. Further study is needed to determine whether these associations are causal and mutable.